Fintech Special Report Part 3: The limits of algos

Fintechs bet on data and algos to know the credit-worthiness of the underserved. However, there's a gap in what the data can reveal because of consumer behaviour. Part 3 of a 4-part series

N S Ramnath

[Fintechs think they have all the data that they need, but they don’t. Because only some of the transactions happen digitally. The rest happen offline. Image from Unsplash]

Big tech firms like Google, Netflix, and Amazon grew rapidly by leveraging data.

The spectacular growth of many big technology firms was fuelled by data. By collecting and analysing data, Big Tech built better products, attracted millions of users and increased sales. For example, Google grew by guiding users to the right information by being a giant data-capturing/crunching machine. Netflix could overtake traditional distribution channels by using data to recommend and produce movies and TV shows. Amazon became one of the world's largest retailers by access to data to recommend products to users, streamline operations, and even launched its own product lines. The general rule in the tech business is that companies with better data and algorithms understand customers better, and are best placed to capture the market.

Fintech companies believe they can crack the hard problem of credit assessment through financial and other data

Fintech companies believe they can leverage data to gain insights into customer behaviour and overtake traditional banking and financial services companies that are burdened by legacy systems.

The fundamental promise of lending startups is that they can identify credit-worthy customers traditionally ignored by the incumbents, often because they are not on regular payrolls. Defaults tend to be higher for this group, which impacts even those who are willing and able to service their loans. The chances of an apple being bad are high. Fintech startups say they can look at financial and other data to identify ‘good apples’ from this basket, thereby expanding the market, revenues and profits.    

However, there are data gaps

Many fintechs are betting on the increasing digitisation of the informal economy. Street vendors, auto-rickshaw drivers and other customer segments that were excluded, now use digital payments apps, generating data which gives a good picture of their cash flows and therefore, their creditworthiness.

"The problem with this approach,” Anuradha Rao, former MD of State Bank of India, says, “is that you think you have all the data that you need, but you don't. Because only some of the transactions happen digitally. The rest happen offline.”

Consider Namma Idly Shop, a popular eatery near Ulsoor Lake in Bangalore. The eatery is bustling in the mornings, with a small team multitasking—preparing food, taking orders, serving customers, managing inventory, and handling billing. Most customers pay using UPI QR codes, and the staff listens for confirmation of payment from the Soundbox. However, some customers prefer to end their meal with a cup of coffee at New Shanti Sagar, just across the road. The bigger eatery has a person dedicated to billing. Often, the same customers who used UPI in the idli shop, pay in cash, creating a gap in the digital records.

The gap, in fact, starts with people who don’t access digital tools in the first place. While digitisation has made life easier for customers who already had access to banking services, the internet and smartphones, the traditionally excluded members have faced significant challenges in accessing these services.

According to a survey by Dvara Research, among those who didn't use any digital payment apps, 40% said it was because they didn't have a smartphone. Besides that, there were other reasons—such as fear of losing money.

While many speak of people leapfrogging to mobile phones and smartphones, the apps are still not intuitive for many, and they need others’ help to install and use them. This in turn has shaped how customers use digital payments apps. For example, many don't do high-value transactions digitally to limit the risk of losing money. In some cases, women are comfortable using it with family and friends, but they are reluctant to use it with strangers, for fear of giving away their phone numbers. While VPA, or virtual payment address, should take care of it, the awareness is low, a sales executive from a payments company says.

Alternative data is still in the experimental stage

Some companies have tried using alternative data, such as rental payments and mobile usage, but there is still scepticism about it. When you crunch data, you will always find correlations between two variables. But, unless you employ scepticism you are likely to burn your fingers, says the CTO of a fintech company.

Banks are open to it, but approach it with caution. “These are still experimental,” says Deepak Sharma, president and chief digital officer at Kotak Mahindra Bank. “For every Rs 100 that a bank disperses, they probably take a bet on 5 or 7, typically for small ticket loans.”

Part of the reason why banks are cautious is that they are concerned about the quality and the source of data. “We don't trust the data that is coming through an intermediary. We want data that is verifiable at source and cannot be tampered with. Otherwise, the cost of verification will be too high and it will not make sense to do business,” a banker said. 

There are more radical approaches to extracting data from a market. One method is to keep the entry barrier low, say, by giving credit cards to everyone who applies, and depending on their repayments, increase, decrease or block credit. It could make business sense in a market that lacks credit information, but it would also forever exclude people who have had the intent to repay the loan but couldn't during a specific period. 

Cracking the market will need more than data

Financial decision-making in India is shaped by a combination of cultural, social, and policy factors like risk aversion, strong savings mentality, social proof, varying financial literacy, and a high trust in government. Indians traditionally prefer low-risk investments like bank deposits or post office saving schemes due to risk aversion, and yet many have fallen for scams en masse, because of social proof. A cultural emphasis on savings and tangible assets like gold impacts choices. Cultural factors also determine why and when people spend. Disparities in literacy and awareness affect decisions, highlighting the need for education programmes. Government policies like financial inclusion initiatives and consumer protection shape financial behaviours. 

There is huge behavioural variance between different groups however one slices and dices the population. One popular way is to divide India into three: the rich, the middle class and the poor. One proxy used to be whether they own a smartphone, a feature phone or no phone at all. But these proxies also break down. There are geographical differences too, not only among different states but also within states, even at a district level. 

All these demand a deep understanding of the market that can only come from feet on the ground. Shriram Finance, for example, relies on references from existing customers, bypassing the need for credit scores. However, this depends on building a relationship with customers, and that can happen only with people on the ground. Not everything can be captured by data, leave alone remotely.    

Eventually, there could be data that captures all these nuances, but the banking and finance industry is not there yet. The data that banks have could be useful, and the account aggregator system, a protocol to share financial information, might help to bring those with specific technologies to extract and use the data. But, Sharma of Kotak Mahindra Bank says there needs to be more conversation between the banks and fintech players. 

Whether fintechs do it in partnerships with banks or alone, understanding the full supply chain is a way forward—like NBFCs have done in the past, but by using tech. 

There’s a tendency among technologists to unbundle activities and then find out ways to improve each of them. But often in the real world, it’s integrated. For example, a senior executive from a milk cooperative said that while financing needs are evident there are many factors at play. Banks are often reluctant to give loans because the government announces write-offs of loans. Meanwhile, private suppliers give loans, and in turn, agree to buy milk at a lower cost. Milk farmers are happy because they at least have money when they need it, even though they might be paying more than the market rate. To break this cycle, there is a need for a deep understanding of the market.

Similarly, extraneous factors can impact household finances significantly. SG Anil Kumar, founder and CEO of Samunnati, a platform that connects farmers and agri-enterprises with the market, says, “We realised that understanding the household in depth is important, but the household does not operate in a vacuum. It also operates in an ecosystem that is usually small and dependent on one or two major economic activities.” For example, in Kumbakonam in Tamil Nadu, the water release in the Mettur dam determines the specific variety of rice that can be cultivated. One variety takes 120 days to grow, while another takes 90 days. This means that if the water release is delayed, the household which grows rice will be affected.

In Karnataka, the Kolar microfinance crisis was caused by volatility in the sericulture value chain. The prices of silk fell, which had a ripple effect on the households that were dependent on sericulture. “We realised that households are not immune to the volatility of the economic activities in their ecosystem. Even if the household is not directly engaged in the activity, volatility in the activity will have an impact on the household. This is why we decided to engage in value chain activities as a way to help households,” says Samunnati’s Anil Kumar.

For startups, it means discarding the mindset that technology alone can solve problems. Technology is just one Lego block. There are others—business models, culture, partnerships, and deep and evolving understanding of human beings and the system. 

In Part 4: In addition, fintech must contend with inherent conflicts with what the regulators want. Fintech’s push for rapid scaling needs to balance with the regulators’ push for financial stability and consumer protection.

Read all 4 parts here.

Was this article useful? Sign up for our daily newsletter below


Login to comment

About the author

N S Ramnath
N S Ramnath

Senior Editor

Founding Fuel

NS Ramnath is a member of the founding team & Lead - Newsroom Innovation at Founding Fuel, and co-author of the book, The Aadhaar Effect. His main interests lie in technology, business, society, and how they interact and influence each other. He writes a regular column on disruptive technologies, and takes regular stock of key news and perspectives from across the world. 

Ram, as everybody calls him, experiments with newer story-telling formats, tailored for the smartphone and social media as well, the outcomes of which he shares with everybody on the team. It then becomes part of a knowledge repository at Founding Fuel and is continuously used to implement and experiment with content formats across all platforms. 

He is also involved with data analysis and visualisation at a startup, How India Lives.

Prior to Founding Fuel, Ramnath was with Forbes India and Economic Times as a business journalist. He has also written for The Hindu, Quartz and Scroll. He has degrees in economics and financial management from Sri Sathya Sai Institute of Higher Learning.

He tweets at @rmnth and spends his spare time reading on philosophy.

Also by me

You might also like